IJRET: International Journal of Research in Engineering and Technology ISSN: 2319-1163

__________________________________________________________________________________________
Volume: 01 Issue: 03 | Nov-2012, Available @ http://www.ijret.org 479
ISOLATED WORD RECOGNITION USING LPC & VECTOR
QUANTIZATION

M. K. Linga Murthy
1
, G.L.N. Murthy
2

1,2
Asst. Professor, ECE, Lakireddy Balireddy College of Engineering, Andhra Pradesh, India,
lingamurthy413@gmail.com, murthyfromtenali@gmail.com

Abstract
Speech recognition is always looked upon as a fascinating field in human computer interaction. It is one of the fundamental steps
towards understanding human recognition and their behavior. This paper explicates the theory and implementation of Speech
recognition. This is a speaker-dependent real time isolated word recognizer. The major logic used was to first obtain the feature
vectors using LPC which was followed by vector quantization. The quantized vectors were then recognized by measuring the
Minimum average distortion.

All Speech Recognition systems contain Two Main Phases, namely Training Phase and Testing Phase. In the Training Phase, the
Features of the words are extracted and during the recognition phase feature matching Takes place. The feature or the template thus
extracted is stored in the data base, during the recognition phase the extracted features are compared with the template in the
database. The features of the words are extracted by using LPC analysis. Vector Quantization is used for generating the code books.
Finally the recognition decision is made based on the matching score. MATLAB will be used to implement this concept to achieve
further understanding.

I ndex Terms: Speech Recognition, LPC, Vector Quantization, and Code Book.
-----------------------------------------------------------------------***-----------------------------------------------------------------------
1. INTRODUCTION
Speech is a natural mode of communication for people. We
learn all the relevant skills during early childhood, without
instruction, and we continue to rely on speech communication
throughout our lives. It comes so naturally to us that we dont
realize how complex a phenomenon speech.

Speech recognition, or more commonly known as automatic
speech recognition (ASR), is the process of interpreting
human speech in a computer. A more technical definition is
given by Jurafsky, where he defines ASR as the building of
system for mapping acoustic signals to a string of words. He
continues by defining automatic speech understanding (ASU)
as extending the goal to producing some sort of understanding
of the sentence.

1.1 Challenges
The general problem of automatic transcription of speech by
any speaker in any environment is still far from solved. But
recent years have seen ASR technology mature to the point
where it is viable in certain limited domains.

1.2 Difficulties
One dimension of variation in speech recognition tasks is the
vocabulary size.
A second dimension of variation is how fluent, natural or
conversational the speech is isolated word recognition, in
which each word is surrounded by some sort of pause, is much
easier than recognizing continuous speech

A third dimension of variation is channel and noise.
Commercial dictation systems, and much laboratory research
in speech recognition, is done with high quality, head mounted
microphones

A final dimension of variation is accent or speaker-class
characteristics. The objective of this paper is to recognize the
isolated words spoken by the speaker. These results are very
useful for implementing the recognition systems. The words or
utterances are recorded by Microphone and are stored in work
space, then processed using MATLAB signal processing
toolbox. It involves pre-emphasis, frame blocking
autocorrelation analysis, LPC Analysis and Vector
quantization.

2. LPC FOR SPEECH RECOGNITION:
There are three basic steps in Speech Recognition, they are
Parameter estimation. ( in which the test pattern
is created )
Parameter Comparison.
Decision making.
IJRET: International Journal of Research in Engineering and Technology ISSN: 2319-1163

__________________________________________________________________________________________
Volume: 01 Issue: 03 | Nov-2012, Available @ http://www.ijret.org 480


Fig2.1 Block diagram for LPC Processor for Speech
Recognition.

2.1 Pre emphasis:
From the speech production model it is known that the speech
undergoes a spectral tilt of -6dB/oct. To counteract this fact a
pre-emphasis filter is used. The main goal of the pre-emphasis
filter is to boost the higher frequencies in order to flatten the
spectrum. Pre emphasis follows a 6 dB per octave rate. This
means that as the frequency doubles, the amplitude increases 6
dB. This is usually done between 300 - 3000 cycles. Pre
emphasis is needed in FM to maintain good signal to noise
ratio. Perhaps the most widely used pre emphasis network is
the fixed first-order system:



2.2 Frame Blocking:
In this step the pre-emphasized speech signal is blocked into
frames of N samples, with adjacent frames being separated by
M samples .Thus frame blocking is done to reduce the mean
squared predication error over a short segment of the speech
wave form. In this step the pre emphasized speech
signal, is blocked into frames of N samples, with
adjacent frames being separated by M samples.



Fig2.2 blocking of speech into overlapping frames.

Typical values for N and M are 256 and 128 when the
sampling rate of the speech is 6.67 kHz. These correspond to
45-msec frames, separated by 15-msec, or a 66.7-Hz frame
rate.

2.3 Windowing:
Here we want to extract spectral features of entire utterance or
conversation, but the spectrum changes very quickly.
Technically, we say that speech is a non-stationary signal,
meaning that its statistical properties are not constant across
time. Instead, we want or extract spectral features from a
small window of speech that characterizes a particular sub-
phone and for which we can make the assumption that the
signal is stationary. this is done by using a window which is
non-zero inside some region and zero elsewhere, running this
window across the speech signal, and extracting the waveform
inside this window. A more common window used in feature
extraction is the Hamming window, which shrinks the values
of the signal toward zero at the window boundaries, avoiding
discontinuities.




2.4 Autocorrelation analysis:
Each frame of windowing signal is next auto correlated to give



Where the highest autocorrelation value, p, is the order of the
LPC analysis. Typically, values of p from 8 to 16 have been
used, with p = 10 being the value used for this systems. A side
benefit of the autocorrelation analysis is that the zeroth
autocorrelation, , is the energy of the frame.

2.5 LPC Analysis:
The next processing step is the LPC analysis, which converts
each frame of P+1 autocorrelations into an LPC parameter set
in which the set might be the LPC coefficients, the reflection
coefficients(PARCOR-co-efficient), the log area ratio
coefficients, or any desired information of the above sets. The
formal method for converting from autocorrelation
coefficients to an LPC parameter set ( for the Autocorrelation
method ) is known as Durbins method

2.6 LPC parameter conversion to Cepstral
coefficients:
A very important LPC parameter set, which can be derived
directly from the LPC coefficients set, is the LPC Cepstral
coefficients, . The recursion method is used. The
Cepstral coefficients, which are the coefficients of the Fourier
transform representation of the log magnitude spectrum, have
been shown to be a more robust, reliable feature set for speech
IJRET: International Journal of Research in Engineering and Technology ISSN: 2319-1163

__________________________________________________________________________________________
Volume: 01 Issue: 03 | Nov-2012, Available @ http://www.ijret.org 481
recognition than the LPC coefficients, the PARCOR
coefficients, or the log area ratio coefficients.

3. VECTOR QUANTIZATION:
Vector quantization is one very efficient source-coding
technique. Vector quantization is a procedure that encodes a
vector of input (e.g., a segment of waveform or a parameter
vector that represents the segment spectrum) into an integer
(index) that is associated with an entry of a collection
(codebook) of reproduction vectors.

Using the basic pattern-recognition approach, the input
(unknown class or word) Speech pattern is next compared
with each class (or word) reference pattern and a measure
(score) of similarity between the unknown pattern and each
reference pattern is calculated.



Fig 4.1 A Vector Quantized based speech recognition system

It is the pattern recognition based approach to speech
recognition, we see that the M codebooks are analogous to M
(sets of) reference patterns (or templates) and the dissimilarity
measure is defined according to Eq. (4.1) and Eq. (4.2), where
no explicit time alignment is required.



The utterance is recognized as class K if



Where is a minimum average distortion.
And is an M average distortion score of all code
books.

4. SYSTEM IMPLEMENTATION:
The training set for the vector quantizer was obtained by
recording the utterances of set isolated words .The words are
recorded for a two different speakers. The recognition
vocabulary consisted of the names (Forward, Back, Left,
Right, and Stop). Here each word is applied for 10 times .The
results obtained are shown in the table below.

Fig 4.1 Comparisons for isolated word recognition system for
two speakers

CONCLUSIONS
Isolated Word Recognition using Linear Predictive Coding
and Vector Quantization provides basic idea for implementing
for Speech Recognition for Isolated Words. The Vector
Quantized based speech recognition is very simple method. In
this method by using LPC analysis extracts the features of
given words & vector quantization is used for feature
matching in Speech Recognition.

In the successful implementation the results were found to be
satisfactory considering less number of training data. The
accuracy of the real time system can be increased significantly
by using an improved speech detection/noise elimination
algorithm.

REFERENCES
[1]. L.R.Rabiner and B.H.Juang, Fundamentals of
speech recognition, Prentice Hall (Signal Processing
series) 1993.
[2]. Richard O.Duda, Peter E.Hart, David
G.Stork,Pattern Classification, John Wiley & Sons
(ASIA) Pte Ltd.
[3]. Y.Linde ,A.Buzo and R.M.Gray, An algorithm for
vector quantizer design ,IEEE Trans .COM-
28,January 1980 .
[4]. Mayukh Bhaowal and kunal chawla , Isolated word
recognition for English language using LPC,VQ &
HMM , students of IIT, Allahabad, India.
[5]. Poonam Bansal , Amita dev and Shail Bala Jain
Automatic speaker Identification using VQ ,
Medwell journals,6(9) : 938 942 ,2007.
[6]. Lawrence R. Rabiner Applications of speech
recognition in the area of Telecommunications,
AT&T Labs Florham Park, New Jersey 07932, 0-
7803-3698-4/97/$10.00 0 1997 IEEE
[7]. L. R. Rabiner, Applications of Voice Processing to
Telecommunications, Proc. IEEE, Vol. 82, No. 4,
pp. 199-228, Feb. 1994.
IJRET: International Journal of Research in Engineering and Technology ISSN: 2319-1163

__________________________________________________________________________________________
Volume: 01 Issue: 03 | Nov-2012, Available @ http://www.ijret.org 482
[8]. L. R. Rabiner and R.W. Schafer Digital processing
of Speech signals, Prentice Hall (Signal Processing
series).
[9]. J.E.Shore and D.K.Burton Discrete Utterance
Speech recognition without Time alignment, IEEE
Transactions on IT, IT 29(4), July 1983, pp. 473
491.
[10]. D. K. Burton, J. T. Buck, and F. Shore, "Parameter
Selection for Isolated Word Recognition Using
Vector Quantization," Proc. ICASSP 84, San Diego,
CA, pp. 9.4.1- 9.4.4, March 1984.
[11]. Douglas oshaughnesy Speech Communication,
Universities press Electrical engineering Series, 2/e
,2001

BIOGRAPHIES:
M. K. Linga Murthy is currently
working as Asst. Professor in ECE in
LBRCE, Mylavaram. He completed
his B.Tech in the year 2001 at SJCET,
Yemmiganur. He completed his
M.Tech in the year 2008 at MITS,
Madanapalle. He has over 06 years of
teaching experience & his research an
area is signal processing.


G.L.N Murthy is currently working as
Asso. Professor in ECE in LBRCE,
Mylavaram. He completed his B.Tech in
JNTU, Anantapu. He completed his
M.Tech in JNTU, Anantapur. He
pursuing his PhD in SVU Tirupathi, he
has over 12 years of teaching experience
& his research an area is signal
processing